Transient Stability Prediction Using algorithm Backpropagation

Current electricity needs are increasing along with the use of electricity in household appliances, offices and so on, so that electricity supply must increase. Request Electricity that continues to increase will result in changes in the condition of the system which is usually caused by a short circuit in the electrical power system. As a result of changes in this condition of system, the system will be change from the old to the new. The short period between the two conditions is called the transitional or transient period. This study aims to determine whether the system is stable or unstable in the event of a disturbance, so an Artificial Neural Network (ANN) is needed to determine the predictions of transient stability. The ANN used in this study uses backpropagation. Backpropagation algorithm is used to compare the performance of each NN system that has a number of hidden layers and different neurons used in performance. Performance can be assessed with several types of parameters, one of them is MSE (Mean Square Error). From several experiments training at random data in backpropagation, the smallest errors were found in 7 hidden layers and each hidden layer had 10 neurons. The data used is data that has never been used in the training phase, the results obtained in this test are prediction targets of transient stability to determine whether the system is stable or unstable in the event of a disturbance short circuit 3 phase. Testing is used there 25%, 50%, 75%, and 100% from 61 data, so that accuracy in testing data shows that backpropagation has achieved good and accurate that is 100%.